Escolar Documentos
Profissional Documentos
Cultura Documentos
ARTICLE
Modeling swelling–shrinkage behavior of compacted expansive
soils during wetting–drying cycles
Gang Wang and Xing Wei
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
Abstract: This paper presents a straightforward approach for modelling volume change behavior of expansive soils during
wetting–drying cycles. The swelling–shrinkage strain of expansive soils induced by cyclic wetting and drying was decomposed
with distinctive physical background into a reversible component, which shows a synchronous change with the cyclic change of
suction, and an irreversible component, which is generated mainly in the early stage of the wetting–drying process. The
mechanisms of the two swelling–shrinkage strain components can be well explained through the double-level structure of
expansive soils and its evolution with mechanical and hydraulic loading. The reversible component originates from the revers-
ible deformation behavior of aggregates, and primarily depends on current suction or water content. The irreversible compo-
nent is associated with the irreversible change of macrostructure, reflecting the difference in soil structures at current state and
the equilibrium state. A practical constitutive model was proposed for compacted expansive clays from a global and phenom-
enological perspective. The model parameters can be calibrated with observed macroscopic deformation behavior without
measuring microstructural parameters. The performance of the presented model was validated by simulating cyclic suction-
controlled tests as well as an alternately soaked and dried test with irregular amplitudes of suctions.
Can. Geotech. J. 2015.52:783-794.
Résumé : Cet article présente une approche directe pour modéliser les variations de volume de sols expansifs durant des cycles
de mouillage–séchage. La déformation causée par le gonflement–retrait des sols expansifs induite par les cycles de mouillage–
séchage a été décomposée, selon des notions physiques, en une composante réversible qui démontre une variation synchrone
avec les variations cycliques de succion, et une composante irréversible qui est générée principalement dans les premières étapes
du processus de mouillage–séchage. Les mécanismes des deux composantes de la déformation en gonflement–retrait peuvent
être bien expliqués grâce à la structure à deux niveaux des sols expansifs et par l’évolution de cette structure avec les charges
mécaniques et hydrauliques. La composante réversible provient de la déformation réversible des agrégats, et dépend principale-
ment de la succion présente et de la teneur en eau. La composante irréversible est associée avec les variations irréversibles
de la macrostructure, reflétant les différences dans les structures du sol à l’état présent et à l’état d’équilibre. Un modèle
constitutif pratique est proposé pour les argiles expansives compactées basé sur une perspective globale et phénomé-
nologique. Les paramètres du modèle peuvent être calibrés avec les déformations macroscopiques observées sans mesurer
les paramètres microstructuraux. La performance du modèle présenté a été validée en simulant des essais de succion
cyclique contrôlés de même d’avec un essai alternant le mouillage et le séchage avec des amplitudes irrégulières de succion.
[Traduit par la Rédaction]
Introduction pansive soils. These tests can be grouped into two categories:
Expansive soils can exhibit large volume change from change in alternately soaked and dried tests (Dif and Bluemel 1991; Day
soil moisture. Under seasonal changes in climatic and environ- 1994; Basma et al. 1996; Fleureau et al. 2002; Tripathy et al. 2002;
mental conditions, expansive soils in situ inevitably experience Tripathy and Subba Rao 2009; Lin and Cerato 2013), and cyclic
cyclic wetting–drying processes and generate swelling–shrinkage suction-controlled tests (Alonso et al. 2005; Nowamooz and
deformation, which results in heave- and subsidence-related cracks Masrouri 2008; Nowamooz et al. 2009; Airò Farulla et al. 2010). In
or deflections in building foundations, buried utilities, highways, alternately soaked and dried tests, only water contents of soil
and airfield pavements, and so on (Chen 1988). Consequently, the specimens are measured directly, therefore the wetting–drying
swelling–shrinkage behavior of expansive soils during wetting– path is usually quantified in terms of volumetric strain versus
drying cycles has been an extensively investigated subject. Fur- water content; while in cyclic suction-controlled tests, soil speci-
thermore, in recent years, the possibility of using compacted mens are allowed to swell or shrink in equilibrium with a prede-
expansive mixtures as engineered barriers or buffers for radioac- termined applied suction, therefore the wetting–drying path is
tive waste disposal (Alonso et al. 2005, 2011) has attracted growing usually characterized by the relationship between volumetric
interest in this subject. strain and suction. Regardless of the difference in cyclic wetting–
A number of laboratory cyclic wetting–drying tests have been drying paths, the following consensuses have been reached from
conducted to estimate the swelling–shrinkage behavior of ex- both categories of tests:
Can. Geotech. J. 52: 783–794 (2015) dx.doi.org/10.1139/cgj-2014-0059 Published at www.nrcresearchpress.com/cgj on 24 October 2014.
784 Can. Geotech. J. Vol. 52, 2015
1. The swelling or shrinkage potential after several wetting– major obstacles to implement this constitutive model into current
drying cycles may reduce or increase significantly compared geotechnical practice; hence, an alternative approach is necessary.
with that in the first cycle. Therefore, the assessment of the vol- The primary objective of the present paper is to develop a
ume change behavior of expansive soils in situ without consid- mathematical description with a relatively small number of easily
ering cyclic seasonal fluctuations leads to unreasonable results. determined parameters for the swelling–shrinkage behavior of
2. Soils reach an equilibrium state after several wetting–drying cy- compacted expansive soils during wetting–drying cycles. The
cles. Once the equilibrium state is reached, the wetting–drying overall philosophy is that the mathematical description and
path becomes fully reversible whether it is in terms of volumetric parameter-determining methodology are developed from a phe-
strain versus water content or in terms of volumetric strain ver- nomenological perspective, without partitioning global defor-
sus suction. mation into microstructure deformation and macrostructure
3. Before the equilibrium state is reached, there exists an irrevers- deformation, although the concept of double-level structure has
ible swelling or shrinkage accumulation as the wetting–drying been adopted to interpret the law and mechanism of the global
cycles go on. Whether the accumulated deformation is expan- deformation phenomena. The presented paper is organized as
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
sion or shrinkage depends on soil properties, initial placement follows. First, a decomposition of the swelling–shrinkage strain
conditions (initial structure), surcharge or confining pressure, due to wetting–drying effects into a reversible component and an
swelling–shrinkage pattern, etc. The magnitude of accumulated irreversible component is introduced based on experimental facts.
shrinkage increases with the increase of applied surcharge or The physical mechanisms of each swelling–shrinkage strain com-
confining pressure; conversely, the magnitude of accumulated ponent are revealed through the double-level structure of expan-
expansion decreases with the increase of applied surcharge or sive clays and its evolution along with mechanical and hydraulic
confining pressure. loading. Based on the physical mechanism, the mathematical de-
An extra and essential feature of constitutive models for expan- scriptions of the two components are presented, which can be
sive soils, which distinguishes them from the constitutive models combined with the Barcelona basic model (BBM) or any other model
for low-plasticity unsaturated soils, is the ability to describe the for unsaturated clays, leading to a complete constitutive model for
swelling–shrinkage behavior during wetting–drying cycles. Gens compacted expansive clays. The determination of the model pa-
and Alonso (1992) and Alonso et al. (1999) presented one of earliest rameters are discussed subsequently in details. Finally, the valid-
elastoplastic models for expansive clays, termed Barcelona expan- ity of the proposed model is confirmed through simulating two
Can. Geotech. J. 2015.52:783-794.
sive model (BExM). The BExM is based on the consideration of two sets of cyclic suction-controlled tests and an alternately wetting
structure levels of expansive clays: the microstructural level at and drying test.
which swelling of active minerals takes place, and the macro-
structural level responsible for major structural rearrangement Two swelling–shrinkage strain components
(Alonso et al. 1999). The formulations of the BExM include defini- The swelling–shrinkage behavior of expansive clays can be well
tions of behavior of the macrostructural level, behavior of the understood in laboratory cyclic suction-controlled tests, in which
microstructural level, and mechanical coupling between the two wetting–drying cycles are performed by cyclic varying applied
structural levels. Within the BExM framework, a complete model suction. Figure 1 shows two typical relationships between volu-
for expansive soils can be established by combining existing metric strain and suction cycle number separately published by
models for each structural level (Mašín 2013). Such models are Alonso et al. (2005) and Nowamooz et al. (2009), where the com-
therefore called double-structure models. Recent advances of pressive strains are considered positive. In both tests, the surcharge
double-structure models include the extension of model formu- pressure remained constant and is less than the preconsolida-
lations to more general stress space or more complex stress paths tion pressure of the specimen at saturated state to ensure the
(Sánchez et al. 2005), and consideration for the coupling effect volumetric strain was induced only by the effect of wetting and
between thermal, mechanical, and hydraulic behavior (Alonso drying. To reduce the test duration, suction was changed in a
et al. 2011; Gens et al. 2011; Della Vecchia et al. 2013; Mašín 2013). single step between the lower and higher values of each half cycle
Beyond the scope of double-structure models, some constitutive in both studies. Airò Farulla et al. (2010) have shown that the size
models suitable for special practical cases were also developed
of applied suction increments has little influence on eventual
for the sake of simplicity, in which the independent behavior of
volumetric strain. As can be seen clearly from Fig. 1, the samples
macrostructure and microstructure of expansive soils were not
swell during wetting phase and shrink during drying phase alter-
considered. For example, Cui et al. (2002) assumed realistically
nately in each cycle, but the volumetric strain generated in the
that the role of macroporosity in swelling of dense compacted
two phases of a cycle is not fully reversible. The samples exhibit
soils may be neglected, and developed a global mechanical model
based on the microstructural model. Similarly, Koliji et al. (2008) and shrinkage (Fig. 1a) or swelling (Fig. 1b) accumulation during cyclic
Najser et al. (2012) described the behavior of double porosity lumpy wetting–drying processes. The rate of the volumetric strain accu-
soils using a model based on the microstructural behavior, with mulation reduces, and the samples tend towards a nearly revers-
phenomenological incorporation of the influence of macrostructure. ible or elastic-like state as the numbers of wetting–drying cycles
The BExM framework successfully explained some essential fea- increase. The majority of the accumulated volumetric strain oc-
tures of expansive soils and received wide acceptance. However, curs during the first wetting–drying cycle. As summarized earlier
double-structure models are often burdened with complexity and in the text, similar phenomena have been confirmed by many
parameters that are difficult to obtain. Unconventional experi- other experimental studies performed by Dif and Bluemel (1991),
mental tests, which are usually difficult to conduct for many lab- Day (1994), Al-Homoud et al. (1995), Basma et al. (1996), Alonso
oratories, are required to quantify the microstructure parameters, et al. (2005), Nowamooz and Masrouri (2008), and Airò Farulla
and a relative large number of parameters are proposed for mi- et al. (2010).
crostructure deformation and coupling function between the two The volumetric strain (swelling–shrinkage strain) due to the
structure levels. There remains a need for a simple and reliable effect of wetting and drying is denoted as vs in this work, since
method to determine the values of these parameters. Although the magnitude of wetting–drying effect is in terms of suction s. As
Alonso et al. (2005) introduced some simplifications of the original demonstrated in Fig. 1, vs can be decomposed into two different
model (Alonso et al. 1999) to facilitate the determination of the mi- components, “a reversible swelling–shrinkage strain component”,
crostructural parameters, the coupling function and corresponding denoted as vs,re, and “an irreversible swelling–shrinkage strain
parameters are still difficult to determine. This has been one of the component”, denoted as vs,ir, namely,
Fig. 1. Decomposition of volumetric strain induced by cyclic wetting and drying: (a) Alonso et al. (2005); (b) Nowamooz et al. (2009).
vs, volumetric strain induced by cyclic wetting and drying; vs,re and vs,ir, reversible and irreversible components of vs, respectively;
v, vertical surcharge pressure.
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
Can. Geotech. J. 2015.52:783-794.
(1) vs ⫽ vs,re ⫹ vs,ir content influence predominantly macroporosity, whereas hydraulic
loading (wetting or drying) influence mainly microporosity, and
also have an effect on macroporosity, depending on the stability
The reversible swelling–shrinkage strain component vs,re always
of macrostructure and the intensity of the wetting–drying effect.
leads to contraction during drying and expansion during wetting,
Figure 3a gives a typical mercury intrusion porosimetry (MIP) test
and shows a synchronous change with the cyclic change of suc-
result published by Monroy et al. (2010), which shows the evolu-
tion. It is further worth noting that the same suction change in
tion of microstructure with wetting. The microporosity increases
each cycle induces nearly the same change of the reversible
with decreasing suction due to swelling of aggregates. The mac-
swelling–shrinkage strain component. This implies that the re-
roporosity, however, remains largely unchanged, except for the
versible swelling–shrinkage strain component mainly depends on
last step of wetting from suction 40 to 0 kPa, in which the porosity
the current suction. The irreversible swelling–shrinkage strain
becomes monomodal, and the double porosity structure is not
component vs,ir may be swelling or shrinkage, and is generated
clear any more. Note that although the porosity is monomodal
mainly in the early stage of the wetting–drying process. The accu-
mulation rate of the irreversible swelling–shrinkage strain compo- upon saturation, the bimodal porosity can be recovered by subse-
nent reduces sharply with increasing number of wetting–drying cycles. quent drying (Simms and Yanful 2001; Cuisinier and Laloui 2004;
Romero et al. 2011). Figure 3b gives a typical result of microstruc-
Physics of two swelling–shrinkage strain ture of compacted kaolin after static compaction to different
static pressures published by Sivakumar et al. (2006). The macro-
components porosity decreases with increasing load, which implies that the
Theoretical background macropores closed up with increasing load. However, the micro-
The soil structure is an important aspect to be considered when porosity slightly increases with increasing compaction effort. A
studying expansive soils. It is well accepted that expansive clays rational interpretation of the latter phenomenon is that the size
have two distinct structure levels, i.e., microstructure level and of a proportion of macropores enters into the micropore-size range
macrostructure level. As illustrated in Fig. 2, clay minerals form as a result of the closure of macropores; therefore, the volume of
individual elementary particle arrangements in a certain config- the intra-aggregate pores possibly remains unchanged. As noted
uration (for example, parallel or flocculent configuration); further by Mašín (2013), the result given by Fig. 3b does not imply that the
elementary particle arrangements join together to make individ- aggregates are undeformable with load. Rather, it implies that the
ual aggregates. The expansive clays are made up of aggregates and aggregate deformation is reversible under mechanical loading be-
other silt or sand grains (sometimes). The microstructure is asso- cause the porosity of the samples compacted to different stresses
ciated with the internal structure of aggregates, whereas the mac- was measured on samples removed from the testing apparatuses,
rostructure is associated with the macroscopic soil skeleton formed and thus on unloaded samples.
by aggregates and other grains. The micropore corresponds to the It is evident that the response of aggregates under mechanical
intra-aggregate pore, whereas the macropore corresponds to the and hydraulic loading is crucial for the deformation behavior of
inter-aggregate (grain) pore. expansive soils and is the key feature distinguishing expansive
Previous studies (Simms and Yanful 2001; Cuisinier and Laloui clays from nonactive clays. The deformation of aggregates is con-
2004; Sivakumar et al. 2006; Lloret and Villar 2007; Miao et al. trolled by physicochemical interactions at particle level and can
2007; Thom et al. 2007; Monroy et al. 2010; Romero et al. 2011) have be assumed dependent mainly on hydraulic loading and indepen-
shown that loading under constant suction or constant water dent of macrostructural effects. Furthermore, the deformation of
Fig. 3. (a) Development of pore-size distribution of compacted London clay with wetting (Monroy et al. 2010); (b) pore-size distribution of
Can. Geotech. J. 2015.52:783-794.
compacted kaolin after static compaction to different static pressures (Sivakumar et al. 2006). ␦enw, change in intruded void ration; ␦(logx),
change of logarithm of entrance pore diameter.
aggregates was usually assumed purely volumetric and reversible macrostructure of expansive soils could be envisaged as a granular-like
under both mechanical and hydraulic loading (e.g., Gens and skeleton formed by individual aggregates. Furthermore, unlike
Alonso 1992). The postulation that the aggregate deformation is sand or silt grains, the aggregates would swell or shrink when
reversible can be confirmed directly by digital image analysis of suction or water content changes. Thus, the deformation of ex-
environmental scanning electronic microscope (ESEM) micro- pansive soils is due to two parts: (i) the rearrangement of the
graphs, in which the volume change of aggregates can be ex- granular-like skeleton formed by the aggregates; and (ii) the self-
tracted. Romero and Simms (2008) and Airò Farulla et al. (2010) deformation of the aggregates. To be consistent with the view that
have reported that the aggregate deformation shows good revers- an aggregate is treated as a discrete body, a “macrostructure void
ibility by this method. Besides the elastic-like behavior of aggre- ratio”, denoted as eM to distinguish from the macrovoid ratio eM
gates under loading, the aggregates with unstable structure may that is more conventional in expansive soil mechanics, was intro-
break down. Katti and Shanmugasundaram (2001) observed that duced to quantify the inter-aggregate porosity for expansive soils.
the increase in swelling and reduction of swelling pressure due to Similar to the void ratio for granular materials, eM is defined as the
swelling in the clay samples result in appreciable reduction in ratio of total inter-aggregate pore volume Vp,M to total apparent
particle size. Thus, the swelling results in the breakdown of some volume of aggregates VA, namely,
unstable aggregates into smaller sized ones. In this study, the
unstable aggregate was treated as an assemblage of smaller stable (2) eM ⫽ Vp,M /VA
aggregates; therefore, the postulation that the aggregate deforma-
tion is reversible is still effective.
If we ignore the internal complexity of aggregates and treat Note that VA includes both solid volume, Vs, and intra-aggregate
each aggregate as a discrete body (like a sand or silt grain), the pore volume, Vp,m, in aggregates, i.e.,
where vs,A is the volumetric strain of aggregates; and eM and eM0
are the current and initial macrostructure void ratios, respec-
tively. The first term at the right-hand side of eq. (3) corresponds
to the apparent deformation due to microstructure change, and
the second term corresponds to the apparent deformation due to
macrostructure change.
Fig. 5. Physical mechanism of irreversible swelling–shrinkage component: (a) open macrostructure; (b) envisaged macrostructure in
equilibrium; (c) closed macrostructure.
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
Model framework
Net volumetric stress p and suction s are adopted as the two Fig. 6. Loading–collapse yield locus and loading–collapse path.
basic stress variables for isotropic stress states. For oedometer
Can. Geotech. J. 2015.52:783-794.
where vp is the mechanical loading induced strain; and vs is the
hydraulic loading induced strain (i.e., the swelling–shrinkage
strain). It should be pointed out that vp can also be induced by a
change in suction s, which may result in stress yielding (wetting
induced collapse), because stress yield locus or yield stress de- where p0 and p∗0 are, respectively, preconsolidation pressures at
pends on suction. For this situation, the name “mechanical load- current and saturated states; pc is a reference stress; is the slope
ing induced strain” still holds valid, and the reason will be of the unloading–reloading line in an e–lnp space and supposedly
explained later. remains constant for both saturated and unsaturated states; (s) is
The formulations of vp, vs,re, and vs,ir will be specified in the the slope of the virgin consolidation line at a suction s; (0) is the
following section to present a complete model for unsaturated slope of the virgin saturated consolidation line; r is the ratio of
expansive clays. The formulations of vs,re and vs,ir are critical for (∞)/(0); and  defines the increasing of stiffness against suction.
reasonable prediction of the swelling–shrinkage behavior. Once the preconsolidation pressure p0 is defined, which serves as
yielding stress, the formulation of vp can be given within the
Model ingredients regime of the plasticity theory as follows:
The formulation of vp can follow any common unsaturated soil
model. The loading–collapse (LC) yielding surface in the BBM is (s) ⫺
e p
employed to combine the suction together with the external stress to (10) dvp ⫽ dvp ⫹ dvp ⫽ dp ⫹ 具d(p ⫺ p0)典
(1 ⫹ e0)p (1 ⫹ e0)p
describe mechanical behavior of unsaturated soils. Figure 6 shows
the LC yielding locus in p–s space, which is defined as follows
e
(Alonso et al. 1990): (11) dvp ⫽ dp
(1 ⫹ e0)p
(8) p0 ⫽ pc 冉冊p∗0
pc
[(0)⫺]/[(s)⫺]
(12) p
dvp ⫽
(s) ⫺
(1 ⫹ e0)p
具d(p ⫺ p0)典
e p
with where dvp and dvp are, respectively, the elastic and plastic parts
of dvp; e0 is the initial void ratio, and its micro and macro parts
(9) (s) ⫽ (0)[(1 ⫺ r)exp(⫺ s) ⫹ r] are not distinguished here for the sake of simplicity; 具 典 is the
Macauley symbol, i.e., 具x典 ⫽ x for x > 0 and 具x典 ⫽ 0 for x ≤ 0. It is Fig. 7. Illustration of cyclic drying–wetting paths and coupling of
clear from Fig. 6 and eq. (8) that both an increase in p and a wetting and LC yielding.
decrease in s may result in a decrease in p0, which leads to plastic
yielding. That is why vp can also be induced by a change in suction.
The following relations are suggested for vs,re, based on the
experimental facts and presented mechanism.
1 ⫹ eM ds
(13) dvs,re ⫽ dvs,A ⫽ cre (1 ⫺ vp ⫺ vs,ir)
1 ⫹ eM0 s ⫹ pat
cre,0
(14) cre ⫽ exp(⫺ ␣1p/pat)
1 ⫹ e0
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
s ⫹ pat
The determination method is discussed in detail as follows.
The value of ⌬vs,re induced by a single-step change in suction
cir,0 p
(16) cir ⫽ ln from an initial value, s1, to a final value, s2, at the constant applied
1 ⫹ e0 pr stress, p, can be obtained by integrating eq. (13):
ⱍ 冉 冊
s2
where cir is the irreversible swelling–shrinkage coefficient; cir,0 is cre s2 ⫹ pat
a material constant; ␣2 defines the decreasing rate of cir along with (18) ⌬vs,re ⫽ ln (1 ⫺ vp ⫺ vs,ir)
s1 1 ⫹ e0 s1 ⫹ pat
the increasing of vs,ir; pr is the critical swelling stress. vs,ir is
contractive when p > pr, whereas vs,ir is expansive when p < pr.
Equations (15) and (16) can well reproduce the dependence of the Equation (18) shows a fully reversible behavior of the volumetric
irreversible swelling–shrinkage strain component on surcharge deformation with the change of suction providing that vp and
pressure and wetting–drying history. In the BExM, this depen- vs,ir remain constant. Given eq. (18) and a suction-change step, cre
dence is represented by involving a p/p0 term in the microstructure– could be determined by fitting the relationship between ⌬vs,re
macrostructure coupling functions fI and fD (Alonso et al. 1999). and the corresponding suction change (s1 to s2). A preliminarily
Figure 7 shows two wetting–drying paths under constant net estimated value of vs,ir can be used in curve fitting. By fitting the
mean stress, indicated as p1 and p2, respectively. The p1 path is relationship between cre and p/pat under different net mean stresses,
located initially inside the LC surface, therefore only vs is gener- cre,0 and ␣1 can be determined.
ated; whereas the p2 path intersects the initial LC surface, where Once the reversible swelling–shrinkage parameters cre,0 and ␣1
suction decrease in the first wetting step would activate the LC are calibrated, the time series of vs,re can be calculated; conse-
quently, the accumulation curves of vs,ir can be extracted from
surface, and thus both vp and vs would be generated. To avoid
the curves of vs (see Fig. 1) for calibrating cir,0, ␣2, and pr. For the
this coupling deformation mechanism and provide convenience
case of shrinkage accumulation (cir > 0, vs,ir > 0), integrating
of data interpretation, the surcharge pressures in most cyclic wet-
eq. (15) along the wetting–drying path from the initial state to the
ting–drying tests were selected with values less than the precon- current state leads to following relationship.
solidation stress values at saturated states.
The p∗0 term in eq. (8) defines the position of the LC surface, and 1
varies in value with the development of plastic strain (hardening (19) vs,ir ⫽ ln(1 ⫹ ␣2cirS)
␣2
law). A coupling hardening law of p∗0 , which can reflect the change
of macrostructure due to wetting–drying effect, is suggested.
with
dp∗0 1 ⫹ e0
共dvp
ⱍ ⱍ
⫹ dvs,ir兲
兺 ln冉 s ⫹ p 冊
p
(17) ⫽ s2 ⫹ pat
p∗0 (0) ⫺
(20) S⫽
1 at
Note that the hardening law here is not identical to that in the
BExM. The total value of vs is employed as an internal variable in where 兺 is the sum of each suction change step during the previ-
the BExM, while only its irreversible component vs,ir is employed ous wetting–drying process; S can be considered as a comprehen-
here. This is consistent with the mechanism of the two swelling– sive variable measuring the experienced wetting–drying effect in
shrinkage strain components because only vs,ir quantifies the history. The terms ␣2 and cir can be determined by fitting the path
rearrangement of macroconfiguration. of vs,ir versus S. For the case of swelling accumulation (cir < 0,
Table 1. List of presented model parameters. Table 2. Model parameters for two expansive clays.
Category Symbol Description Category Symbol Alonso et al. Nowamooz et al.
LC yielding v∗ or p0∗ Initial saturated preconsolidation stress BBMⴱ v∗ (kPa) 650 600
pc Reference stress pc (kPa) 0.008 200
(0) Saturated virgin compression index (0) 0.25 0.19
Elastic compression or rebound index 0.045 0.04
 Parameter that defines the curvature of  (MPa−1) 0.0544 0.60
the LC yield curve r 0.85 0.70
r Parameter that defines the limiting value Presented cre,0/(1 + e0) 0.02 0.012
of the compression index for high model ␣1 0.001 0.002
suctions cir,0/(1 + e0) 0.007 0.041
Wetting and cre,0 Irreversible swelling–shrinkage coefficient pr (kPa) 90 65
drying when p = 0 ␣2 25 70
BExMⴱ m
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
(23) e
vm ⫽ 冉 m
1 ⫹ e0 冊冋 d共p ⫹ s兲
p⫹s
册 by the presented model and BExM, respectively. Material param-
eters of the two models are given in Table 2. Simulation results of
both models are also plotted in Fig. 8 for comparison. The pre-
sented model performs well in predicting the two tests at vertical
where pvM is the plastic macrostructural volumetric strain; vme
is net stresses of 196 and 396 kPa. However, the comparison is not so
the microstructural volumetric strain; m is the elastic compress- good for the test at net vertical stress of 98 kPa because this test
ibility index of microstructure; and fI and fD are the micro–macro has an initial large swelling strain (about 9%), which has an
coupling functions. The formulations of fI and fD can be obtained opposite direction to the successive strain accumulation. For
by interpolating the experimental data points. The sigmoidal func- simplicity, the presented model postulates that the irreversible
tions as follows are commonly adopted (e.g., Alonso et al. 2005). swelling–shrinkage strain component is gradually monotoni-
cally accumulated; obviously, this is not in accordance with this
冋冉 冊册
f I1 ⫺ f I2 v f I1 ⫹ f I2 special case. Although the presented model failed in reproducing
(24) fI ⫽ tan⫺1 KI ∗ ⫺ XI ⫹ the intermediate strain path, the predicted eventual strain mag-
v0 2
nitude was in accordance with the test result.
Fig. 8. Simulation of cyclic suction-controlled tests of Alonso et al. (2005): (a) v = 396 kPa; (b) v = 196 kPa; (c) v = 98 kPa.
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
Can. Geotech. J. 2015.52:783-794.
Simulation of cyclic suction-controlled tests reported by the clayey soil has the following characteristics: liquid limit is
Nowamooz et al. (2009) 85.6%; plasticity index is 31.9%; specific gravity is 2.6; and clay
Nowamooz et al. (2009) performed experimental studies on a content (<2 mm) is 72%. The preconsolidation pressure of the soil
natural French expansive clayey soil. The experimental sample of at saturated state is about 600 kPa. Three cyclic wetting–drying
Fig. 9. Simulation of cyclic suction-controlled tests of Nowamooz et al. (2009): (a) v = 20 kPa; (b) v = 40 kPa; (c) v = 60 kPa.
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
Can. Geotech. J. 2015.52:783-794.
tests were performed at vertical net stresses of 20, 40, and 60 kPa, the volumetric strains converged towards an equilibrium stage. A
respectively, with suction change between a low value of 0 and a high completely reversible behavior was reached for the test of v = 60 kPa,
value of 2 MPa. The test procedure is in common with that of Alonso whereas one or two additional cycles would have been necessary for
et al. (2005). The volumetric strains of the three samples are shown in tests of v = 20 and 40 kPa to reach completely reversible stages.
Fig. 9 where the swelling strains are considered positive. The samples The predictions of the proposed model and BExM are drawn in
showed a swelling accumulation as a result of successive cycles, and Fig. 9 against test results. With one group of parameters, the pro-
Fig. 10. Simulation of wetting–drying test involving full shrinkage and partial shrinkage of Tripathy and Subba Rao (2009). s, suction;
w, water content; v, vertical surcharge pressure.
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
posed model shows good prediction capability for all the three reversible component and an irreversible component with distinc-
Can. Geotech. J. 2015.52:783-794.
tests as well as a soaked and dried wetting–drying test with irreg- Koliji, A., Vulliet, L., and Laloui, L. 2008. New basis for the constitutive modelling of
ular amplitudes of cyclic water content changes. aggregated soils. Acta Geotechnica, 3(1): 61–69. doi:10.1007/s11440-007-
0052-x.
Lin, B., and Cerato, A.B. 2013. Hysteretic soil water characteristics and cyclic
Acknowledgement swell-shrink paths of compacted expansive soils. Bulletin of Engineering
The present study is financially supported by the National Natural Geology and the Environment, 72: 61–70. doi:10.1007/s10064-012-0450-7.
Science Foundation of China (No. 51209179, No. 2010CB732103). Lloret, A., and Villar, M.V. 2007. Advances on the knowledge of the thermo-
hydro-mechanical behaviour of heavily compacted “FEBEX” bentonite. Physics
References and Chemistry of the Earth, Parts A/B/C, 32: 701–715. doi:10.1016/j.pce.2006.
03.002.
Airò Farulla, C., Ferrari, A., and Romero, E. 2010. Volume change behaviour of a Mašín, D. 2013. Double structure hydromechanical coupling formalism and a
compacted scaly clay during cyclic suction changes. Canadian Geotechnical
model for unsaturated expansive clays. Engineering Geology, 165: 73–88.
Journal, 47(6): 688–703. doi:10.1139/T09-138.
doi:10.1016/j.enggeo.2013.05.026.
Al-Homoud, A.S., Basma, A.A., Husein Malkawi, A.I., and Al Bashabsheh, M.A.
Miao, L., Houston, S.L., Cui, Y., and Yuan, J. 2007. Relationship between soil
1995. Cyclic swelling behavior of clays. Journal of Geotechnical Engineering,
structure and mechanical behavior for an expansive unsaturated clay. Cana-
121(7): 562–565. doi:10.1061/(ASCE)0733-9410(1995)121:7(562).
dian Geotechnical Journal, 44(2): 126–137. doi:10.1139/t06-108.
Alonso, E.E., Gens, A., and Josa, A. 1990. A constitutive model for partially satu-
Downloaded from www.nrcresearchpress.com by 47.8.225.124 on 03/28/19. For personal use only.
rated soils. Géotechnique, 40(3): 405–430. doi:10.1680/geot.1990.40.3.405. Monroy, R., Zdravkovic, L., and Ridley, A. 2010. Evolution of microstructure in
compacted London clay during wetting and loading. Géotechnique, 60(2):
Alonso, E.E., Vaunat, J., and Gens, A. 1999. Modelling the mechanical behaviour
of expansive clays. Engineering Geology, 54: 173–183. doi:10.1016/S0013-7952 105–119. doi:10.1680/geot.8.P.125.
(99)00079-4. Najser, J., Mašín, D., and Boháč, J. 2012. Numerical modelling of lumpy clay
Alonso, E.E., Romero, E., Hoffmann, C., and García-Escudero, E. 2005. Expansive landfill. International Journal for Numerical and Analytical Methods in Geo-
bentonite-sand mixtures in cyclic controlled-suction drying and wetting. En- mechanics, 36(1): 17–35. doi:10.1002/nag.990.
gineering Geology, 81: 213–226. doi:10.1016/j.enggeo.2005.06.009. Nowamooz, H., and Masrouri, F. 2008. Hydromechanical behaviour of an expan-
Alonso, E.E., Romero, E., and Hoffmann, C. 2011. Hydromechanical behaviour of sive bentonite/silt mixture in cyclic suction-controlled drying and wetting
compacted granular expansive mixtures: experimental and constitutive tests. Engineering Geology, 101: 154–164. doi:10.1016/j.enggeo.2008.04.011.
study. Géotechnique, 61(4): 329–344. doi:10.1680/geot.2011.61.4.329. Nowamooz, H., Mrad, M., Abdallah, A., and Masrouri, F. 2009. Experimental and
Basma, A.A., Al-Homoud, A.S., Husein Malkawi, A.I., and Al-Bashabsheh, M.A. numerical studies of the hydromechanical behaviour of a natural unsatu-
1996. Swelling-shrinkage behavior of natural expansive clays. Applied Clay rated swelling soil. Canadian Geotechnical Journal, 46(4): 393–410. doi:10.
Science, 11(2–4): 211–227. doi:10.1016/S0169-1317(96)00009-9. 1139/T08-127.
Chen, F.H. 1988. Foundations on expansive soils. Elsevier, New York. Romero, E., and Simms, P.H. 2008. Microstructure investigation in unsaturated
Cui, Y.J., Yahia-Aissa, M., and Delage, P. 2002. A model for the volume change soils: A review with special attention to contribution of mercury intrusion
behavior of heavily compacted swelling clays. Engineering Geology, 64: 233– porosimetry and environmental scanning electron microscopy. Geotechni-
Can. Geotech. J. 2015.52:783-794.